Accurate evaluation and quantitative characterization of deep coal porefracture structure (PFS) is crucial for enhancing coalbed methane recovery efficiency. A novel approach combining transverse relaxation time and nuclear magnetic resonance (NMR) cryoporometry was employed to obtain a comprehensive pore size distribution by using NMR equipment alone. The relationship between the PFS and reservoir quality index was formulated. Two NMR permeability prediction models are further proposed for deep coals, providing a "one-stop" PFS permeability evaluation. The results show that it is practical to determine the surface relaxivity and obtain the pore size distribution by combining NMR cryoporometry tests with the surface relaxivity for coals from the mine of Pingdingshan, China, estimated to be approximately 4.63 μm/s. Seepage pore porosity, fractal dimension, and connectivity can be used to quantitatively characterize the differences in the PFS among coal samples and predict their influence on permeability. Coal samples with superior reservoir quality tended to display seepage pores characterized by high porosity, low heterogeneity, and enhanced connectivity. The proposed NMR permeability prediction models can accurately predict deep coal permeability, providing a rapid, nondestructive, and effective reservoir quality evaluation method.